Fraud Detection Tool

The Fraud Detection Tool, created as a dissertation project, is a web-based application designed to identify potential fraud in transaction data uploaded as CSV files.

Built with Flask, it leverages both supervised (Random Forest) and unsupervised (Isolation Forest) machine learning models to analyze datasets, automatically detecting whether a fraud label column (e.g., "Class" or "Fraud") is present to select the appropriate model. Users can upload transaction data, view detected fraud cases with associated probabilities, and download results as a CSV file. For supervised models, the tool provides detailed performance metrics, including precision, recall, F1-score, and accuracy, to evaluate model effectiveness. The intuitive interface ensures accessibility for non-technical users, while the robust backend handles complex data processing and model training, demonstrating a comprehensive solution for fraud detection in various domains as part of academic research.